Re: [v6ops] Lion/Snow Leaopard side-by-side on an IPv6 enabled ADSL2+ home line..

[Philip Homburg <[e-mail]>]

In your letter dated Fri, 5 Aug 2011 08:14:15 -0400 you wrote:
>Don't defend HE as being the best solution either.    HE is of pretty =
>narrow applicability, really.    The document doesn't overstate its =
>applicability, and the algorithm described in the document is not bad =
>for web browsers, which is why I have no objection to it.   (and I'm =
>happy to see IETF work in this area)   But the algorithm isn't going to =
>work well for apps in general.
>
>Also, my understanding is that Apple doesn't force programmers to use =
>the clever API.     Programmers can still (and sometimes should) use =
>simpler APIs and do their own style of path selection that better fits =
>the needs of their specific application.

I could be wrong, but I think I read somewhere that Apple does reorder the
results of getaddrinfo. If that is true then application that use
the basic socket extensions would end up with HE whether they want it or
not.

>Architecturally speaking, putting a global preference policy based on =
>address type into every host is a very dubious idea.  It cannot work =
>well in general.

Note that the HE draft is smart enough to say that the address selection
policy should be followed unless there is evidence that a certain address
or protocol doesn't work all that well.

So there is the possibility of local overrides.